ABSTRACT Immune checkpoint inhibitors (ICIs) have revolutionized cancer treatment but clinical responses remain limited due to primary and/or acquired ICI resistance. Preliminary studies suggest the gut microbiome is an independent, novel modulator of systemic and intra-tumoral responses to ICIs. Published studies, primarily in melanoma and non-small cell lung cancer (NSCLC), reported a diverse gut microbiome and a few bacterial species associated with tumor responses to ICIs. However, existing studies are limited by an oversimplified classification of responder status and lack of longitudinal analysis. Moreover, specific bacteria or bacterial communities putatively helpful or harmful to ICI responses have largely been inconsistent across study populations and tumor types with only limited correlations with immune or mutational biomarkers. To address these inconsistencies, we reanalyzed raw 16S rRNA amplicon and whole genome sequencing (WGS) fecal data from five published studies (n=303, pre-treatment stools only) using a high dimensional computational approach and found that our predictive index underperformed in NSCLC, the leading cause of cancer death, compared to melanoma. Thus, to further refine this index and identify mechanisms of action for the microbiome in NSCLC ICI therapy, herein, we propose detailed metabolite (plasma, fecal) and metagenomic (fecal, WGS) longitudinal studies of our NSCLC cohort (n=108). We will focus our studies on patients with a durable response to ICIs, comparing them to those with primary resistance (disease progression within 6 months) or acquired resistance (disease progression after an initial response >6 months). Microbiota metabolite production, with the capacity to impact local and systemic immune responses, is proposed as a ‘common pathway’ mechanism promoting anti-tumor responses. For example, both recent literature and our preliminary data in transplantable tumor murine models suggest inosine and short chain fatty acids (SCFA) are microbial modifiers of the tumor microenvironment. Further, a preliminary study found SCFAs enriched in fecal sample from NSCLC patients with durable response to ICIs (n=11). We hypothesize that specific bacterial species and/or communities promote production of SCFAs and/or inosine in the gut and plasma of NSCLC patients who exhibit durable responses to ICI therapy whereas these are absent in patients who exhibit primary and/or ultimately develop acquired resistance. To test our hypothesis, we propose to identify fecal and/or plasma metabolites (SA1) and/or microbial taxa (SA2) that correlate with durable response or ICI resistance in NSCLC patients. We will further test our hypothesis by defining the intra-tumoral immune cell response to our patient cohorts in syngeneic murine tumor models using single cell RNAseq and paired T cell receptor sequencing to define T cell repertoires (SA3). Our multidimensional approach, with both cross-sectional and longitudinal translational an...